720 keys USB PC DIY keyboard

[alain]

Hi guys,

At the moment, I know nothing more about this subject than the fact I have the desire to build a programmable 720 keys PC usb mechanical keyboard (15 rows x 48 columns ) with Cherry MX Black switches. Could you help me by pointing me in the right technical direction ?
Thanks very much.

[Mechboards]

Wow. That's big.

First thing - Go here - http://www.keyboard-layout-editor.com/ and create your desired layout. This can be used to create the mapping file. Then you could take the Raw Data to http://builder.swillkb.com/ to create a case and top plate.

Next is wiring. I have no clue if any controller works up to that size.

[SpAmRaY]

I'm curious to know the usage scenario?

[3K]

While I think no option to realize this keyboard will be easy, the easiest would probably to take several smaller keyboards and make them work together.

For example you take 6 of these: http://i.imgur.com/qfrSb3V.jpg (The model / brand usually goes by name Tipro)
Rotate them by 90 degree and put them side by side. This would give you an keyboard with layout 16 x (8 + 8 + 8 + 8 + 8 + 8 ) = 16 x 48.
I'd say that's close enough.

Edit: ALSO Tipros usually already have MX Blacks installed. ALSO you could just leave a row without function / empty whatever.

[alain]

It's for keyboard shortcuts in music production: right now I'm using 4 side by side keyboards. I'd like to replace them with a single one at some point.

[TalkingTree]

The only way to do it that I can think off the top of my head is by daisy chaining a long array of shift registers.

[Puddsy]

handwire it

[alain]

sure, I'll handwire the keys, but to what ?

[TalkingTree]

sure, I'll handwire the keys, but to what ?

To a breakout board with 63 input/output pad which, as far as I know, doesn't exist. You could surely handwire it but that would be a massive job, besides, you would still need interface expanders or shift registers to build such a large matrix.
Futhermore, know that your product would be around 90 * 30 cm large, which would make either your PCB and/or mounting plate very expensive.

[SpAmRaY]

I know this doesn't have nearly enough keys but OP might find it useful

https://www.reddit.com/r/mechmarket/comments/6407uc/gb_squared/

[Skull_Angel]

Have you thought about using multiple layers to reduce the number of keys needed? I know some programmables currently have up to 5 layers (mostly 60% boards). Would be a good way to reduce the size if you can handle memorizing the layers.

[dantan]

TBH I don't see how your desk can accommodate 6 keyboards. Mine can barely accommodate one full size board.

[dorkvader]

TBH I don't see how your desk can accommodate 6 keyboards. Mine can barely accommodate one full size board.

I have 5 on my desk right now. 1 SSK, 1 infinity, one F122 that I'm programming, and an ergodox, there's even a G80-1800 on the floor connected to my NAS. there's room for more. Some people have big desks.

Anyway regarding the topic at hand, creating a KB this big would be an interesting challenge. From a hardware perspective: not too difficult, you can get a controller with a lot of pins and then use IO expanders or mux to get all the keys. If you can afford it, getting a PCB printed with diodes on it would save a lot of time wiring. Even several PCBs, one for each IO expander would save a lot of effort. Large one-off PCBs are expensive though.

I don't know much about DAW or whatever else you use for music production, but the KB firmware might be a difficult thing. Using the firmware we currently have, you would end up having to map keys to do combinations, since the project far exceeds the number of usable scancodes, so you would have to do something like "press physical key 700, send keycodes for press and release for [A][Q]" for example. I believe this is possible suing current KB controller firmware, but it'll depends on how they work with IO expanders / mux, etc.

fortunately used cherry black switches are cheap and plentiful, though used browns might be cheaper since they were made into the POS keyboards that go on ebay for like $30

If you do go with PCB, you could get a piece of material (plastic, metal, wood even) to use as a base, and do PCB mount switches and then screw the PCBs into the backing. If you go with hand-wire, then you will need a very large custom plate manufactured which might be expensive.

In any case this is an expensive project. Still it's entirely possible depending on what your objectives are. For example, how do you use all of your keyboards now? if you press "a" on one of them, that's usually treated as the same "a" as on another keyboard, making it less than useful for many cases. I don't know how your system works, but if you can program it to take key combinations like the above mentioned [A][Q] to map to a specific command it could work.

[tp4tissue]

I dont' understand how you can remember what key does what out of all those keys..

do you have stickers on them?

[ErgoMacros]

I'd attack this by:

• Splitting the field of keys into 4 parts (PCBs), each with 180 keys
• Each part gets 15 rows x 12 column arrays
• To scan this many keys I'd offload the scanning task. Put a micro controller on each PCB
• To scan that many keys you need 15+12 = 27 pins on the processor.
• BTW the GH-122 (search Geekhack) supports about 150 keys in a 7 rows x 24 columns configuration. That's 31 processor pins of a Teensy++ 2.0 being used. More than the 27 that you need
• The PCBs would be something a bit larger than 11.25 x 9 inches. Maybe 12.5 x 36" for all 4?
• The processor on each PCB handles the scanning and reports the results back to a master controller that *maybe* has to track state, and will handle the USB communication.
• Each board communicates via a serial interface to the master. (Like Ergodox and others do)
• The PCBs can be identical (economies of scale). The additional controller is only populated on the master
• You can simplify the firmware downloading by making the controller on each PCB use it's USB port
• You can start small, 1 board, prove it out, than add the others as desired. No need to stop at 4 if you've got the desk space and budget.

Fun project... you can never have too many keys.

[dorkvader]

I'd attack this by:

• Splitting the field of keys into 4 parts (PCBs), each with 180 keys
• Each part gets 15 rows x 12 column arrays
• To scan this many keys I'd offload the scanning task. Put a micro controller on each PCB
• To scan that many keys you need 15+12 = 27 pins on the processor.
• BTW the GH-122 (search Geekhack) supports about 150 keys in a 7 rows x 24 columns configuration. That's 31 processor pins of a Teensy++ 2.0 being used. More than the 27 that you need
• The PCBs would be something a bit larger than 11.25 x 9 inches. Maybe 12.5 x 36" for all 4?
• The processor on each PCB handles the scanning and reports the results back to a master controller that *maybe* has to track state, and will handle the USB communication.
• Each board communicates via a serial interface to the master. (Like Ergodox and others do)
• The PCBs can be identical (economies of scale). The additional controller is only populated on the master
• You can simplify the firmware downloading by making the controller on each PCB use it's USB port
• You can start small, 1 board, prove it out, than add the others as desired. No need to stop at 4 if you've got the desk space and budget.

Fun project... you can never have too many keys.

Having a T++ on every daughterboard is kind of a waste. What you want are like what you said before with an I2C expander. but I haven't looked up if there are any that have support for a matrix that large. You could also use a serial to parallel adapter like the frosty flake, etc replacement costar controller that bpiphany made, but again you might run out of pins still.

Of course the easiest way is to just whack a teensy on each board and report four keyboards to the computer, which I think will work because it's similar to what he has now. Less elegant though.

Alain, have you considered asking for advice on deskthourity too? it can't hurt to see whats out there and get some other opinions on it. I think step 0 should be to more clearly define your problem and goals, which will make it much easier for us to help.

[ErgoMacros]

dorkvader, I agree the Teensy on each board is overkill, and if I were going to mass produce these I'd be more conservative with the computing resources.
I (a software guy, much more than hardware) was just just going for my most known technology. At less than $30US per, Teensy is cheep compared to the time it takes to develop new code for your own "satellite" processors, working on re-flashing them.

Like the idea of just building the 4 pieces and treating each as a separate keyboard. Eliminates the need for board to board communications. As long as that satisfies his requirements.

[alain]

I focus on accelerating the workflow as much as possible. Right now, I use 420 different keys:



My various destination music softwares are at ease with 87 keys of a normal keyboard (no PRINTSCREEN and such for example). With combinations of modifying keys added,

1. normal keys
2. CTRL + normal keys
3. SHIFT + normal keys
4. ALT + normal keys
5. CTRL + SHIFT + normal keys
6. CTRL + ALT + normal keys
7. ALT + SHIFT + normal keys

Give or take a few (in example excluding ones like ALT+F4) that's 87 * 7 = 609 keys commands. I complete this with launching specific executables and urls.

I went for 720 because it's not practical for me to exceed 90x30cm at the moment. But if I redesigned my desk*, had the budget and the time, I wouldn't mind 2000 or 3000 keys. The effectiveness of pressing a single key instead of doing ten different things to obtain the same result is very addictive. Sure it takes time to set up, but your work efficiency improves immensly afterwords, at least that's my personnal experience

Thanks again for all the technical suggestions. I'm building other devices at the same time, so I won't jump right in, but all the ideas are very appreciated for my permanently ongoing autodidaxia.

*speaking of redesigning desks, if you know a motorized sit & stand support system that can withstand real heavy piano keyboards, screens, monitors and such, I'm interested!

Also, following 3K's wonderful suggestion, I emailed Tipro, they were very quick, kind and pro in their answer, recommending me their existing keyboards if that's a solo endeavor, as well as saying they could design such a keyboard if I ordered at least several hundreds of them. I am not a Captain of Industry as we speak, so I won't be doing that, but if hundreds of users around the globe find this big keyboard interesting, Tipro will probably find it a worthy investment of their resources to produce such a keyboard. The pure DIY route is also on my mind.

[Targa-TV]

This sounds colossal. Good luck man.
I feel like the finished product should be printed as a poster (and glued to my door).

[_rubik]

Where is the law of diminishing returns on this. I mean, we can only get our boards so small before they begin to inhibit our work flow. I feel the same happens when they get too large. I mean, if you get 2000 keys as you mentioned -- even with 720 -- you can't possibly efficiently navigate all your keys and remember each and every function.

[alain]

you can't possibly efficiently

I am using that amount of functions daily anyways. The problem is not remembering what they are and where they are, but wasting precious life seconds pointing and mouse clicking thru endless GUI based sub-menus. Another way of explaining it is to say that as far as I'm concerned, goals don't become smaller without these hardware tools, just time-wise borderline impossible to reach. Memory-wise, sure, none of it is for the faint of heart, but what is ? If you so choose, you're anyway faced with the need to remember thousands upon thousands of relevant facts, dates, techniques, processes, instruments, chords, modes, functions in softwares, books, records, people and so on. It's a life choice. So remembering 500+ keyboards shortcuts is nothing, especially when they are grouped by theme, assigning different values to the same set of variables and displaying icons you've thoughtfully designed and printed yourself over the course of several years. Maybe it's not for you but it works extremely well for some

[ander]

Sounds awesome.

One can't help thinking about IBM's early attempts at a Japanese-language board:





At a mere 200-something keys, your board will dwarf it, but it's a step in the right direction.

[daaawooo]

I failed at my Ergodox in the last half year. Missing to many keys. 5 or even 10 layers don't work for me, I need reel dedicated keys (VS key-combos) - but not 720, maybe something between 130-150 will just right for me.
Your project sounds great, please keep us informed.

[_rubik]

Maybe it's not for you but it works extremely well for some

Aint that the truth. Either way. I'm very eager to see where this takes you. Make sure to post the final product.

[ErgoMacros]

...I need reel dedicated keys (VS key-combos) - but not 720, maybe something between 130-150 will just right for me.
...

There's an interesting project I'm following, the GH-122 board, case, etc. Here: https://geekhack.org/index.php?topic=88063.0

Goes up to 150 keys. About 50+ macro/function keys, programable, of course. Ron did a very similar board a year a go. May not be to late to get in on the prototype round. Certainly time for the production.

Assembly required.

[Data]

you can't possibly efficiently

I am using that amount of functions daily anyways. The problem is not remembering what they are and where they are, but wasting precious life seconds pointing and mouse clicking thru endless GUI based sub-menus. Another way of explaining it is to say that as far as I'm concerned, goals don't become smaller without these hardware tools, just time-wise borderline impossible to reach. Memory-wise, sure, none of it is for the faint of heart, but what is ? If you so choose, you're anyway faced with the need to remember thousands upon thousands of relevant facts, dates, techniques, processes, instruments, chords, modes, functions in softwares, books, records, people and so on. It's a life choice. So remembering 500+ keyboards shortcuts is nothing, especially when they are grouped by theme, assigning different values to the same set of variables and displaying icons you've thoughtfully designed and printed yourself over the course of several years. Maybe it's not for you but it works extremely well for some

If you group your commands/macros very carefully on the keyboard they could effectively replace a deeply-nested menu system and that has a lot of potential for improving workflow.  Have you planned out your keymap yet?

[Parak]

The kiibohd firmware is capable of this - for example HaaTa did a proof of concept a while back where he daisy chained 5 ergodox halves to each other, with lcd backlight and everything. It's easy to run up against usb power limitations if everything is backlit or has lcds, but with some careful planning it should be possible. If I recall correctly the mk20dx256vlh7 would support (at least) a 14x10 key matrix without an LCD, and without a backlight that can go up to 18x10/14x14 or thereabouts. You'd be able to do 4 boards chained to each other like that, ending with one usb connector to the computer.

Huge caveat: You're probably going to need to learn how to make your own pcbs and work with the firmware and stuff.

Disclaimer just in case: Am part of Input Club.

[happylacquer]

just get some BCR2000's op?

But honestly i am curious what software you are using, do you have 600~ shortcuts all in one DAW?

[TalkingTree]

One thing that you could do is to daisy chain PCBs and connect them to a controller (with extended I/O).
This way you can modularly expand or shrink your keyboard and, most of all, reduce the production costs.

This should give you a vague idea of how to do it. Predispose a PCB with, say, 5 x 5 switches and a jumper on each border end to connect to the adjacent PCB.

Excuse the awful artwork.

[Severe]

Why tho

[__red__]

So, in the interests of full disclosure I built a full-size functional replica of this from cherry MX black keys:
http://collections.nmmusd.org/Keyboards/JankoPiano/4168UprightpianocloseupofJankokeyboardLG.jpg

I also own one of these:
https://intervalawareness.files.wordpress.com/2014/12/product_wh.png?w=640

My answer to your question comes down to one simple question.  Do you want the input of your keyboard to be HID key-presses (HID doesn't support that many 'key types' so you'll be programming "chording" constantly) or do you want the device to output midi events.

I built a 4x4 key (staggered) keyboard module with LED in 2012 which would tile in both dimensions allowing arbitrary sized keyboard arrays.  Unfortunately I miscalculated the dimensions so had to trim hundreds of keycaps by hand so they'd fit :-(

The modules fit in the 'standard' Chinese PCB size which means that getting them made is really, REALLY cheap.

I wouldn't start with that design today (it's five years old and I've learned a lot since) but you're welcome to poll me for advice.

[ErgoMacros]

alain,
Correct me if I'm wrong, but I'm getting the impression that somewhere along the line you're mapping a keypress to a specific menu, sub-menu, or sub-sub-menu.

Those cascading menu item options get mapped (through what?) to individual keys. So (for example) keyboard "A" might map 2 dozen keys to, say, Ctrl-Alt-A through Ctrl-Alt-Z. Those keys map to the most frequent/useful menus for sound application "X"

So you're avoiding endless mouse-ing or Down, Down, Down, Right, Down-ing.

Yes?

If not,
  * what are you doing now?
  * do you want to replicate that, just changing the form-factor of the switches?
  * or better yet, what is the problem you're trying to solve?
  * what OS are you using?
  * What software are you mapping it with?
  * Is your vision hat every key has one function? or are you, for example, willing to accept 1 "meta" key to double the number of functions per key. (maybe even a foot pedal?)

I think you've got a community ready and willing to help, but the goal is still fuzzy (at least to me).

Thanks

[ander]

Why fool around though? How about a quantum-based keyboard that can create an infinite number of keys, as fast as you can visualize them? Dr. Seuss alluded to something like this in his 1953 film "The 5000 Fingers of Dr. T":

[ErgoMacros]

OK, here's another option, but pre-built, not DIY.

G8663400EUADAA - Cherry SPOS G86-6340 POS Keyboard
    http://www.barcodegiant.com/cherry/part-g8663400euadaa.htm?aw&adtype=pla&gclid=CMKGrODeoNQCFdRyfgod970OzQ

144 key Position Matrix in 14" form factor (14.17 x 7.79 x 1.49 inches) for universal key function assignment. 8 rows x 18 columns. Fully Programmable and Relegendable Keyboard. Option for Unlimited Programmability of all Keys in up to 3 Layers.
About $110 each at nwegg.com and others.

Five (5) keyboards at $110 each, $550 total, 720 keys with nothing to design, fabricate, build, test, or program. (Perhaps.)
3 year warranty included.

[alain]

Thanks to all of you, really ErgoMacros, I possess the POS Keyboard you mentioned. Unfortunately, if you want delay times and sophisticated macros, they need to stay on the software side, which won't distinguish two or more copies of the same model (I asked the Cherry guys a few years ago). I am sticking with cheap keyboards (got two more) until I have more money and time to experiment.

[Phenix]

Do yourself a favor, BUY some POS Tipro keyboards.

[ErgoMacros]

OK, here's another option... 306 keys, one micro processor.

From our friends at reddit:
https://www.reddit.com/r/MechanicalKeyboards/comments/6wj1ri/the_big_ass_macropad_a_306_key_18x17_macropad/

18 x 17 array of keys, but No key rollover.

[ErgoMacros]

And yet another one... still not 720, but 3 of these 15 x 15 (225) gets you to 675:
   https://geekhack.org/index.php?topic=95231.0;topicseen

[ErgoMacros]

OK, another large pad. 15x15 = 225 keys. Three of them would get you to 625 keys.
It's created as a Scrabble board.
https://geekhack.org/index.php?topic=95231.0;topicseen

About to start a group buy.